The international road map of semiconductors (ITRS) lists 2D materials as possible future materials for electronic devices [ITRS]. Among them, the semiconducting transition metal dichalcogenides (TMDC) like MoS2 or WS2 are the most promising ones. As the miniaturization of electronic devices continues and the sub-5 nm gate limit is reached, direct source-to-drain tunneling will become a problem in Si devices with conventional architecture. 2D TMDC as channel material for FET might overcome this issue in digital CMOS. The larger carrier effective masses of most TMDC result in reduced direct source-to-drain tunneling, while they also yield a large density of states and hence an increased ballistic current in this extreme dimensions [Fiori]. Additionally, the lower in-plane dielectric constants of TMDC enable a better vertical electrostatic control over the channel [Desai]. Still, TMDC will have to compete against other approaches, e.g. multiple-gate transistors [Schwiertz]. Beyond that, other properties of TMDC turn them interesting for future devices, such as their strong absorbance across the visible spectrum for optoelectronic devices [Lotsch]. The first photodetectors, electromluminescent p-n diodes and photovoltaic cells have already been shown [Lopez, Cheng, Tsai]. The excellent mechanical properties additionally enable the realization of different kinds of flexible devices [Schwiertz]. And last but not least, the integration of 2D materials into conventional 3D semiconductor concepts would be another application. For example, the strain originating from the large difference in lattice constants in a 3D heterostructure is often a challenge. Inserting 2D materials might be a solution. Due to the fully terminated surface of a TMDC monolayer, the binding to another material is of van-der-Waals type. Hence, TMDC can be stacked with materials with huge differences in lattice constants without leading to considerable strain or to interface trap densities [Vogel]. First devices with a combination of 2D and 3D materials have already been investigated [Krishna, Lee]. Large-scale fabrication of TMDC is still a challenge to be overcome. Up to now, the most frequently used techniques are either exfoliation of natural layered crystals or the growth via chemical vapor deposition (CVD). Exfoliation is a very time-consuming process with relatively low yield and reproducibility. In CVD, different gaseous precursors decompose on a substrate and react with each other. Thermally evaporated MoO3 and S are commonly used as precursors for the fabrication of MoS2 monolayers via CVD. This technique allows the deposition of monolayered crystallites with a size in the range of micrometers [Dumcenco]. The process is carried out in small experimental reactors, and a uniform deposition on a whole wafer is very challenging. The use of metalorganic precursors is one possibility to enter an industrial scale of fabrication. The development of III/V and II/VI semiconductors has shown that metalorganic chemical vapor deposition (MOCVD) makes controllable and reproducible processes feasible, which are easily scalable and hence suitable for large-area deposition. Additionally, the reactors are well-developed, and a future integration of 2D materials and other semiconductors is within reach. First publications show very good results on the deposition of MoS2 and WSe2 via MOCVD [Kang, Eichfeld]. But despite these works and first simulations of the growth kinetics [Nie], little is known about the growth mechanism. For this reason, we have started to investigate the deposition of 2D MoS2 on an AIXTRON MOCVD reactor.The experiments are carried out in a horizontal hot-wall MOCVD reactor in a 10 × 2 inch configuration. Molybdenum hexacarbonyl (MCO) and Di-tert-butyl sulfide (DTBS) are used as Mo and S sources, respectively. In preliminary experiments, the carrier gas based transport of the precursors into the reactor and possible growth conditions are tested. Their results are used to develop an initial growth process. This process leads to a uniform, wafer-scale deposition of MoS2 on various substrate types such as sapphire (0001), Si (111), and AlN and GaN templates on sapphire substrates. The deposited films mainly consists of MoS2 bilayers and exhibit a very high initial nucleation density. Further investigations of the influence of the growth temperature, the carrier gas composition and the pretreatment of the substrates are carried out. With optimization of these growth parameters, a crystal growth process closer to thermodynamical equilibrium can be achieved. This results in the formation of triangular crystals on sapphire substrates which are also reported from CVD processes and exhibit higher crystal quality [Dumcenco]. Additional experiments are conducted to investigate the nucleation of the films and to further tune nucleation density and lateral growth rate in order to deposit wafer-scale monolayered MoS2 films.
Read full abstract